trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 10.0,
  "global_step": 2430,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 2.06,
      "learning_rate": 3.971193415637861e-05,
      "loss": 2.8803,
      "step": 500
    },
    {
      "epoch": 4.12,
      "learning_rate": 2.9423868312757202e-05,
      "loss": 2.3089,
      "step": 1000
    },
    {
      "epoch": 6.17,
      "learning_rate": 1.91358024691358e-05,
      "loss": 2.116,
      "step": 1500
    },
    {
      "epoch": 8.23,
      "learning_rate": 8.847736625514404e-06,
      "loss": 2.0249,
      "step": 2000
    },
    {
      "epoch": 10.0,
      "step": 2430,
      "total_flos": 2.2677509529759744e+16,
      "train_loss": 0.35158030819991,
      "train_runtime": 504.822,
      "train_samples_per_second": 1230.275,
      "train_steps_per_second": 4.814
    }
  ],
  "max_steps": 2430,
  "num_train_epochs": 10,
  "total_flos": 2.2677509529759744e+16,
  "trial_name": null,
  "trial_params": null
}